Feature-index-based similar shape retrieval

A new feature-index-based shape similarity-based image retrieval technique, which is capable of handling images of rigid as well as articulated objects, is proposed and can also retrieve images of overlapping and touching objects that satisfy the query constraints.

Abstract

Efficient retrieval of images containing matching (or similar) shapes is an important problem in image database management. A few shape similarity-based image retrieval techniques have been reported in the literature. Most of these techniques are designed to deal only with images of rigid objects and cannot easily handle shape similarity-based retrieval of images of articulated objects (i.e., objects with one or more joints with movable components). Some of these techniques are incapable of handling images of overlapping and touching objects. In this paper, a new feature-index-based shape similarity-based image retrieval technique, which is capable of handling images of rigid as well as articulated objects, is proposed. The proposed technique can also retrieve images of overlapping and touching objects that satisfy the query constraints. In this approach structural components of a shape called features are represented as points in a multidimensional space. Any multidimensional point access index structure can be used to organize shape features. Given a query shape or image, a query shape feature is selected and similar (or matching) shape features belonging to images in the database are found by searching the feature index structure. If the query shape and a stored shape containing any of the retrieved matching features are found to be similar (or satisfy similarity constraints), the stored shape is included in the response to the query.